support single tissue download

Author: WagnerNils

splicemap/main.py

@@ -70,21 +70,37 @@ def complete_url(tissue, psi):
     ]
 }
 
+def check_tissue_in_url(url, tissues):
+    in_url = False
+    for tissue in tissues:
+        if tissue in url:
+            in_url = True
+    return in_url
+
+def _download(url, splicemap_dir):
+    r = requests.get(url)
+    fname = Path(splicemap_dir)
+    fname.mkdir(exist_ok=True)
+    fname = fname / \
+        unquote(Path(url).name).replace('?download=1', '')
+    print(fname)
+    with open(fname, 'wb') as fd:
+        fd.write(r.content)
+
 @click.command()
 @click.option('--version', help='SpliceMap version (currently SpliceMaps from gtex_v8 (hg38) supported)')
 @click.option('--splicemap_dir', help='Path to download SpliceMaps')
-def splicemap_download(version, splicemap_dir):
+@click.option('--tissues', multiple=True, help='List of tissue names to download')
+def splicemap_download(version, splicemap_dir, tissues=None):
 
     if version not in splicemap_url:
         raise(f'Version {version} is not supported.')
 
     print('Downloading SpliceMaps...')
 
     for url in tqdm(splicemap_url[version]):
-        r = requests.get(url)
-        fname = Path(splicemap_dir)
-        fname.mkdir(exist_ok = True)
-        fname = fname / unquote(Path(url).name).replace('?download=1', '')
-        
-        with open(fname, 'wb') as fd:
-            fd.write(r.content)
+        if tissues:
+            if check_tissue_in_url(url, tissues):
+                _download(url, splicemap_dir)
+        else:
+            _download(url, splicemap_dir)

tests/test_download.py

@@ -1,14 +1,31 @@
 from click.testing import CliRunner
 from splicemap.splice_map import SpliceMap
-from splicemap.main import splicemap_download
-
+from splicemap.main import splicemap_download, gtex_v8_tissues
+import os
 
 def tests_download_splicemap(tmp_path):
     runner = CliRunner()
 
-    result = runner.invoke(splicemap_download, f'--version _test --splicemap_dir {str(tmp_path)}')
+    result = runner.invoke(splicemap_download, f'--version gtex_v8 --splicemap_dir {str(tmp_path)}')
     assert result.exit_code == 0
-
-    sm = SpliceMap.read_csv(str(tmp_path))
-    assert sm.df.shape[0] > 0
 
+    for tissue in gtex_v8_tissues:
+        for psi in ['psi5', 'psi3']:
+            sm = SpliceMap.read_csv(os.path.join(
+                    tmp_path, 
+                    f'{tissue}_splicemap_{psi}_method=kn_event_filter=median_cutoff.csv.gz'))
+            assert sm.df.shape[0] > 0
+    
+def tests_download_splicemap_single_tissue(tmp_path):
+    runner = CliRunner()
+
+    result = runner.invoke(
+        splicemap_download, f'--version gtex_v8 --splicemap_dir {str(tmp_path)} --tissues Whole_Blood --tissues Lung')
+    assert result.exit_code == 0
+
+    for tissue in ['Whole_Blood', 'Lung']:
+        for psi in ['psi5', 'psi3']:
+            sm = SpliceMap.read_csv(os.path.join(
+                    tmp_path, 
+                    f'{tissue}_splicemap_{psi}_method=kn_event_filter=median_cutoff.csv.gz'))
+            assert sm.df.shape[0] > 0